EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Estimating out-of-bank discharge uncertainties using a hydrodynamic model and nationally available datasets

Gemma Coxon, Robert Milsom, and Jeff Neal
Gemma Coxon et al.
  • University of Bristol, Geographical Sciences, Bristol, United Kingdom of Great Britain and Northern Ireland (

Robust predictions and forecasts of flood risks and hazards are reliant on accurate estimates of stream flow data.  However, the stage-discharge relationship is subject to substantial uncertainties from a range of error sources, particularly for out-of-bank flows where measurements are scarce and flows are often extrapolated.  Hydraulic modelling can be used to produce more reliable stage–discharge relationships beyond the range of observed measurements, but, these methods are often data intensive requiring topographical, bathymetric, calibration data etc. restricting their use across large samples of gauges.    

In this study, we present an automatable framework that can estimate out-of-bank discharge uncertainty using a hydrodynamic model and readily available national datasets.  The framework utilises LiDAR data, in-bank stage-discharge measurements and gauged river flows to calibrate a 1D/2D hydrodynamic model (LISFLOOD-FP) of a river reach and make predictions of river flow and rating curve uncertainty beyond bankfull.  A particularly novel aspect of this framework is the use of national LiDAR datasets of water surface elevation returns to estimate the bathymetry and friction in the channel using an inversion solver. 

The framework was applied to produce models of two gauged river reaches in the UK, the River Severn at Montford in Shropshire, and the River Tweed at Norham in Northumberland. Bathymetry estimates were consistent with observations, considering that the channel was simplified to rectangular below the LiDAR water surface, while Manning’s channel friction estimates were between 0.03 and 0.035. The model predictions showed a close fit to the official rating curve and out-of-bank stage-discharge measurements, with the model-predicted uncertainty bounds able to contain 89.5% and 100% of the out-of-bank flow measurements for Montford and Norham respectively. This holds promising results for quantifying out-of-bank discharge uncertainties across large samples of catchments to enable robust national flood risk assessment.

How to cite: Coxon, G., Milsom, R., and Neal, J.: Estimating out-of-bank discharge uncertainties using a hydrodynamic model and nationally available datasets, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-3066,, 2020


Display file

Comments on the display

AC: Author Comment | CC: Community Comment | Report abuse

displays version 1 – uploaded on 03 May 2020
  • CC1: Comment on EGU2020-3066, Maik Renner, 05 May 2020

    Hi Gemma,

    this is very important and practically relevant work!

    I am surprised that your results show a very good relation of the observed gaugings and the predicted ones?

    A technical question is if your inversion can separate the uncertainties of the unknown river profiles and the Mannings roughness? 

    Did you use a DGM for the floodplains for the 1D2D hydrodynamic modelling?

    Thank you,


  • AC1: Comment on EGU2020-3066, Gemma Coxon, 07 May 2020

    Hi Maik

    Thanks for the nice questions.

    In response:

    1. We were pleasantly surprised too!
    2. In our view the roughness and bathymetry estimate are part of a set, one only exists with the other. If you consider the two distributions independently then you would overestimate the uncertainty when sampling from either. 
    3. We assume here you mean a digital elevation model? We used a 5m LiDAR DEM sourced from the Environment Agency.

    Best Wishes,